Submitted to: Remote Sensing and Modeling Applications for Natural Resource Management
Publication Type: Abstract Only
Publication Acceptance Date: 3/13/2002
Publication Date: N/A
Citation: Interpretive Summary:
Technical Abstract: Widespread implementation of precision agriculture will require methods for efficiently and economically characterizing variations in soil properties and other factors that affect crop yields. In this research the potential of using airborne hyperspectral images to estimate within-field soil variability was investigated. Bare soil images were acquired using a prism grating pushbroom scanner (RDACS H3) in April 2000 and May 2001 for a central Missouri experimental field in a minimum-tillage corn-soybean rotation. Images consisted of 120 bands from 471 to 828 nm with a spatial resolution of 1 m. Data were converted to reflectance using chemically-treated reference tarps with eight known reflectance levels. Geometric distortions of the pushbroom sensor images caused by aircraft attitude changes during image acquisition were corrected with a rubber sheeting transformation. Various levels of spatial aggregation were investigated on sub-field areas to determine the optimum pixel size for image analysis. Image data were related to soil properties (organic matter, cation exchange capacity, P, K, Mg, Ca, and pH) obtained by grid sampling on a 30-m spacing and to bulk soil electrical conductivity (ECa) obtained with an in-field sensor. Simple correlation, multiple regression, partial least squares regression, principal component analysis, and canonical correlation analysis were used to relate hyperspectral data to field-measured soil properties. Relationships were also developed between soil properties and Landsat TM-like bands derived by spectral aggregation of the hyperspectral bands. Reflectance in the blue wavelengths was most highly related to grid-sampled soil properties and to soil ECa.